基于相关匹配及双模式模板更新的跟踪方法  

Target Tracking Method Based on Blocks and Two-Model Template Updating

作　　者：徐克虎[1] 李科[1] 张波[1] 

机构地区：[1]装甲兵工程学院控制工程系,北京100072

出　　处：《计算机测量与控制》2012年第4期1058-1060,共3页Computer Measurement &Control

基　　金：国防预研基金项目(9140A09050708JB3503)

摘　　要：运动目标被遮挡表现为一部分子块区域与模板的失配,从而影响目标跟踪效果;为解决该问题,提出了一种改进的相关跟踪方法和双模式模板更新策略;通过自适应阈值调整和分块相关运算对MCD算法进行改进,在遮挡发生时判别被遮挡子块,及时调整模板更新策略,由Kalman滤波结果与匹配结果的线性加权和得出目标最终位置;实验结果表明:该方法能有效抑制目标局部像素变化引起的跟踪漂移,且对遮挡具有一定的鲁棒性。When the moving object is occluded, it means that some blocks area of the target matching falsely, which will affect the result of the target tracking. A method is proposed to solve this problem in complex scene by combining an improved correlation tracking algorithm with two--model template updating strategy. This paper proposes an improved MCD correlation tracking arithmetic by adaptive threshold adjusting and blocks correlation matching. It adjusts model template strategy based on target＇s motion state. The results of Kalman filtering and correlation matching are done with linear weight method to get the final position of the target. The experimental result shows that the presented method can prevent tracking excursion caused by part pixels of target changing, and it is robust to occlusion.

关 键 词：自适应阈值 分块相关 双模式模板更新 遮挡处理 KALMAN滤波 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]